templately / modules / full-site-import / Parsers / WXR_Parser_SimpleXML.php

WXR_Parser_SimpleXML.php in Templately – Elementor & Gutenberg Template Library: 6500+ Free & Pro Ready Templates And Cloud! 3.8.0, at modules/full-site-import/Parsers/WXR_Parser_SimpleXML.php

281 lines 8.5 KB

Raw Download Zip

1	<?php
2	namespace Templately\Modules\FullSiteImport\Parsers;
3
4
5	use Templately\Modules\FullSiteImport\Utils\Utils;
6	use WP_Error;
7
8	if ( ! defined( 'ABSPATH' ) ) {
9	exit; // Exit if accessed directly.
10	}
11
12	/**
13	* WordPress extended RSS file parser implementations,
14	* Originally made by WordPress part of WordPress/Importer.
15	* https://plugins.trac.wordpress.org/browser/wordpress-importer/trunk/parsers/class-wxr-parser-simplexml.php
16	*
17	* What was done (by Elementor):
18	* Reformat of the code.
19	* Removed variable '$internal_errors'.
20	* Changed text domain.
21	*/
22
23	/**
24	* WXR Parser that makes use of the SimpleXML PHP extension.
25	*/
26	class WXR_Parser_SimpleXML {
27
28	/**
29	* @param string $file
30	*
31	* @return array\|WP_Error
32	*/
33	public function parse( $file ) {
34	$authors = [];
35	$posts = [];
36	$categories = [];
37	$tags = [];
38	$terms = [];
39
40	libxml_use_internal_errors( true );
41
42	$dom = new \DOMDocument();
43	$old_value = null;
44
45	$libxml_disable_entity_loader_exists = function_exists( 'libxml_disable_entity_loader' );
46
47	if ( $libxml_disable_entity_loader_exists ) {
48	$old_value = @libxml_disable_entity_loader( true ); // phpcs:ignore Generic.PHP.DeprecatedFunctions.Deprecated
49	}
50
51	$success = $dom->loadXML( Utils::file_get_contents( $file ) );
52
53	if ( $libxml_disable_entity_loader_exists && ! is_null( $old_value ) ) {
54	@libxml_disable_entity_loader( $old_value ); // phpcs:ignore Generic.PHP.DeprecatedFunctions.Deprecated
55	}
56
57	if ( ! $success \|\| isset( $dom->doctype ) ) {
58	return new WP_Error( 'SimpleXML_parse_error', esc_html__( 'There was an error when reading this WXR file', 'elementor' ), libxml_get_errors() );
59	}
60
61	$xml = simplexml_import_dom( $dom );
62	unset( $dom );
63
64	// Halt if loading produces an error.
65	if ( ! $xml ) {
66	return new WP_Error( 'SimpleXML_parse_error', esc_html__( 'There was an error when reading this WXR file', 'elementor' ), libxml_get_errors() );
67	}
68
69	$wxr_version = $xml->xpath( '/rss/channel/wp:wxr_version' );
70	if ( ! $wxr_version ) {
71	return new WP_Error( 'WXR_parse_error', esc_html__( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'elementor' ) );
72	}
73
74	$wxr_version = (string) trim( $wxr_version[0] );
75	// Confirm that we are dealing with the correct file format.
76	if ( ! preg_match( '/^\d+\.\d+$/', $wxr_version ) ) {
77	return new WP_Error( 'WXR_parse_error', esc_html__( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'elementor' ) );
78	}
79
80	$base_url = $xml->xpath( '/rss/channel/wp:base_site_url' );
81	$base_url = (string) trim( $base_url[0] ?? '' );
82
83	$base_blog_url = $xml->xpath( '/rss/channel/wp:base_blog_url' );
84	if ( $base_blog_url ) {
85	$base_blog_url = (string) trim( $base_blog_url[0] );
86	} else {
87	$base_blog_url = $base_url;
88	}
89
90	$page_on_front = $xml->xpath( '/rss/channel/wp:page_on_front' );
91
92	if ( $page_on_front ) {
93	$page_on_front = (int) $page_on_front[0];
94	}
95
96	$namespaces = $xml->getDocNamespaces();
97	if ( ! isset( $namespaces['wp'] ) ) {
98	$namespaces['wp'] = 'http://wordpress.org/export/1.1/';
99	}
100	if ( ! isset( $namespaces['excerpt'] ) ) {
101	$namespaces['excerpt'] = 'http://wordpress.org/export/1.1/excerpt/';
102	}
103
104	// Grab authors.
105	foreach ( $xml->xpath( '/rss/channel/wp:author' ) as $author_arr ) {
106	$a = $author_arr->children( $namespaces['wp'] );
107	$login = (string) $a->author_login;
108	$authors[ $login ] = [
109	'author_id' => (int) $a->author_id,
110	'author_login' => $login,
111	'author_email' => (string) $a->author_email,
112	'author_display_name' => (string) $a->author_display_name,
113	'author_first_name' => (string) $a->author_first_name,
114	'author_last_name' => (string) $a->author_last_name,
115	];
116	}
117
118	// Grab cats, tags and terms.
119	foreach ( $xml->xpath( '/rss/channel/wp:category' ) as $term_arr ) {
120	$t = $term_arr->children( $namespaces['wp'] );
121	$category = [
122	'term_id' => (int) $t->term_id,
123	'category_nicename' => (string) $t->category_nicename,
124	'category_parent' => (string) $t->category_parent,
125	'cat_name' => (string) $t->cat_name,
126	'category_description' => (string) $t->category_description,
127	];
128
129	foreach ( $t->termmeta as $meta ) {
130	$category['termmeta'][] = [
131	'key' => (string) $meta->meta_key,
132	'value' => (string) $meta->meta_value,
133	];
134	}
135
136	$categories[] = $category;
137	}
138
139	foreach ( $xml->xpath( '/rss/channel/wp:tag' ) as $term_arr ) {
140	$t = $term_arr->children( $namespaces['wp'] );
141	$tag = [
142	'term_id' => (int) $t->term_id,
143	'tag_slug' => (string) $t->tag_slug,
144	'tag_name' => (string) $t->tag_name,
145	'tag_description' => (string) $t->tag_description,
146	];
147
148	foreach ( $t->termmeta as $meta ) {
149	$tag['termmeta'][] = [
150	'key' => (string) $meta->meta_key,
151	'value' => (string) $meta->meta_value,
152	];
153	}
154
155	$tags[] = $tag;
156	}
157
158	foreach ( $xml->xpath( '/rss/channel/wp:term' ) as $term_arr ) {
159	$t = $term_arr->children( $namespaces['wp'] );
160	$term = [
161	'term_id' => (int) $t->term_id,
162	'term_taxonomy' => (string) $t->term_taxonomy,
163	'slug' => (string) $t->term_slug,
164	'term_parent' => (string) $t->term_parent,
165	'term_name' => (string) $t->term_name,
166	'term_description' => (string) $t->term_description,
167	];
168
169	foreach ( $t->termmeta as $meta ) {
170	$term['termmeta'][] = [
171	'key' => (string) $meta->meta_key,
172	'value' => (string) $meta->meta_value,
173	];
174	}
175
176	$terms[] = $term;
177	}
178
179	// Grab posts.
180	foreach ( $xml->channel->item as $item ) {
181	$post = [
182	'post_title' => (string) $item->title,
183	'guid' => (string) $item->guid,
184	];
185
186	$dc = $item->children( 'http://purl.org/dc/elements/1.1/' );
187	$post['post_author'] = (string) $dc->creator;
188
189	$content = $item->children( 'http://purl.org/rss/1.0/modules/content/' );
190	$excerpt = $item->children( $namespaces['excerpt'] );
191	$post['post_content'] = (string) $content->encoded;
192	$post['post_excerpt'] = (string) $excerpt->encoded;
193
194	$wp = $item->children( $namespaces['wp'] );
195	$post['post_id'] = (int) $wp->post_id;
196	$post['post_date'] = (string) $wp->post_date;
197	$post['post_date_gmt'] = (string) $wp->post_date_gmt;
198	$post['comment_status'] = (string) $wp->comment_status;
199	$post['ping_status'] = (string) $wp->ping_status;
200	$post['post_name'] = (string) $wp->post_name;
201	$post['status'] = (string) $wp->status;
202	$post['post_parent'] = (int) $wp->post_parent;
203	$post['menu_order'] = (int) $wp->menu_order;
204	$post['post_type'] = (string) $wp->post_type;
205	$post['post_password'] = (string) $wp->post_password;
206	$post['is_sticky'] = (int) $wp->is_sticky;
207
208	if ( isset( $wp->attachment_url ) ) {
209	$post['attachment_url'] = (string) $wp->attachment_url;
210	}
211
212	// Extract attachment_type field
213	if ( isset( $wp->attachment_type ) ) {
214	$post['attachment_type'] = (string) $wp->attachment_type;
215	} else {
216	$post['attachment_type'] = null;
217	}
218
219	foreach ( $item->category as $c ) {
220	$att = $c->attributes();
221	if ( isset( $att['nicename'] ) ) {
222	$post['terms'][] = [
223	'name' => (string) $c,
224	'slug' => (string) $att['nicename'],
225	'domain' => (string) $att['domain'],
226	];
227	}
228	}
229
230	foreach ( $wp->postmeta as $meta ) {
231	$post['postmeta'][] = [
232	'key' => (string) $meta->meta_key,
233	'value' => (string) $meta->meta_value,
234	];
235	}
236
237	foreach ( $wp->comment as $comment ) {
238	$meta = [];
239	if ( isset( $comment->commentmeta ) ) {
240	foreach ( $comment->commentmeta as $m ) {
241	$meta[] = [
242	'key' => (string) $m->meta_key,
243	'value' => (string) $m->meta_value,
244	];
245	}
246	}
247
248	$post['comments'][] = [
249	'comment_id' => (int) $comment->comment_id,
250	'comment_author' => (string) $comment->comment_author,
251	'comment_author_email' => (string) $comment->comment_author_email,
252	'comment_author_IP' => (string) $comment->comment_author_IP,
253	'comment_author_url' => (string) $comment->comment_author_url,
254	'comment_date' => (string) $comment->comment_date,
255	'comment_date_gmt' => (string) $comment->comment_date_gmt,
256	'comment_content' => (string) $comment->comment_content,
257	'comment_approved' => (string) $comment->comment_approved,
258	'comment_type' => (string) $comment->comment_type,
259	'comment_parent' => (string) $comment->comment_parent,
260	'comment_user_id' => (int) $comment->comment_user_id,
261	'commentmeta' => $meta,
262	];
263	}
264
265	$posts[] = $post;
266	}
267
268	return [
269	'authors' => $authors,
270	'posts' => $posts,
271	'categories' => $categories,
272	'tags' => $tags,
273	'terms' => $terms,
274	'base_url' => $base_url,
275	'base_blog_url' => $base_blog_url,
276	'page_on_front' => $page_on_front,
277	'version' => $wxr_version,
278	];
279	}
280	}
281