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

WXR_Parser_XML.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_XML.php

370 lines 7.7 KB

Raw Download Zip

1	<?php
2	namespace Templately\Modules\FullSiteImport\Parsers;
3
4	use Elementor\Utils;
5	use WP_Error;
6
7	if ( ! defined( 'ABSPATH' ) ) {
8	exit; // Exit if accessed directly.
9	}
10
11	/**
12	* WordPress extended RSS file parser implementations,
13	* Originally made by WordPress part of WordPress/Importer.
14	* https://plugins.trac.wordpress.org/browser/wordpress-importer/trunk/parsers/class-wxr-parser-xml.php
15	*
16	* What was done (by Elementor):
17	* Reformat of the code.
18	* Added PHPDOC.
19	* Changed text domain.
20	* Added clear() method.
21	* Added undeclared class properties.
22	* Changed methods visibility.
23	*/
24
25	/**
26	* WXR Parser that makes use of the XML Parser PHP extension.
27	*/
28	class WXR_Parser_XML {
29	private static $wp_tags = [
30	'wp:post_id',
31	'wp:post_date',
32	'wp:post_date_gmt',
33	'wp:comment_status',
34	'wp:ping_status',
35	'wp:attachment_url',
36	'wp:attachment_type',
37	'wp:status',
38	'wp:post_name',
39	'wp:post_parent',
40	'wp:menu_order',
41	'wp:post_type',
42	'wp:post_password',
43	'wp:is_sticky',
44	'wp:term_id',
45	'wp:category_nicename',
46	'wp:category_parent',
47	'wp:cat_name',
48	'wp:category_description',
49	'wp:tag_slug',
50	'wp:tag_name',
51	'wp:tag_description',
52	'wp:term_taxonomy',
53	'wp:term_parent',
54	'wp:term_name',
55	'wp:term_description',
56	'wp:author_id',
57	'wp:author_login',
58	'wp:author_email',
59	'wp:author_display_name',
60	'wp:author_first_name',
61	'wp:author_last_name',
62	];
63
64	private static $wp_sub_tags = [
65	'wp:comment_id',
66	'wp:comment_author',
67	'wp:comment_author_email',
68	'wp:comment_author_url',
69	'wp:comment_author_IP',
70	'wp:comment_date',
71	'wp:comment_date_gmt',
72	'wp:comment_content',
73	'wp:comment_approved',
74	'wp:comment_type',
75	'wp:comment_parent',
76	'wp:comment_user_id',
77	];
78
79	/**
80	* @var string
81	*/
82	private $wxr_version;
83
84	/**
85	* @var string
86	*/
87	private $cdata;
88
89	/**
90	* @var array
91	*/
92	private $data;
93
94	/**
95	* @var array
96	*/
97	private $sub_data;
98
99	/**
100	* @var boolean
101	*/
102	private $in_post;
103
104	/**
105	* @var boolean
106	*/
107	private $in_tag;
108
109	/**
110	* @var boolean
111	*/
112	private $in_sub_tag;
113
114	/**
115	* @var array
116	*/
117	private $authors;
118
119	/**
120	* @var array
121	*/
122	private $posts;
123
124	/**
125	* @var array
126	*/
127	private $term;
128
129	/**
130	* @var array
131	*/
132	private $category;
133
134	/**
135	* @var array
136	*/
137	private $tag;
138
139	/**
140	* @var string
141	*/
142	private $base_url;
143
144	/**
145	* @var string
146	*/
147	private $base_blog_url;
148
149	/**
150	* @param string $file
151	*
152	* @return array\|WP_Error
153	*/
154	public function parse( $file ) {
155	$this->clear();
156
157	$xml = xml_parser_create( 'UTF-8' );
158	xml_parser_set_option( $xml, XML_OPTION_SKIP_WHITE, 1 );
159	xml_parser_set_option( $xml, XML_OPTION_CASE_FOLDING, 0 );
160	xml_set_object( $xml, $this );
161
162	xml_set_character_data_handler( $xml, function ( $parser, $cdata ) {
163	$this->cdata( $cdata );
164	} );
165
166	$tag_open_callback = function ( $parse, $tag, $attr ) {
167	$this->tag_open( $tag, $attr );
168	};
169
170	$tag_close_callback = function ( $parser, $tag ) {
171	$this->tag_close( $tag );
172	};
173
174	xml_set_element_handler( $xml, $tag_open_callback, $tag_close_callback );
175
176	if ( ! xml_parse( $xml, Utils::file_get_contents( $file ), true ) ) {
177	$current_line = xml_get_current_line_number( $xml );
178	$current_column = xml_get_current_column_number( $xml );
179	$error_code = xml_get_error_code( $xml );
180	$error_string = xml_error_string( $error_code );
181
182	return new WP_Error( 'XML_parse_error', 'There was an error when reading this WXR file', [
183	$current_line,
184	$current_column,
185	$error_string,
186	] );
187	}
188	xml_parser_free( $xml );
189
190	if ( ! preg_match( '/^\d+\.\d+$/', $this->wxr_version ) ) {
191	return new WP_Error( 'WXR_parse_error', esc_html__( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'elementor' ) );
192	}
193
194	return array(
195	'authors' => $this->authors,
196	'posts' => $this->posts,
197	'categories' => $this->category,
198	'tags' => $this->tag,
199	'terms' => $this->term,
200	'base_url' => $this->base_url,
201	'base_blog_url' => $this->base_blog_url,
202	'version' => $this->wxr_version,
203	);
204	}
205
206	private function tag_open( $tag, $attr ) {
207	if ( in_array( $tag, self::$wp_tags ) ) {
208	$this->in_tag = substr( $tag, 3 );
209
210	return;
211	}
212
213	if ( in_array( $tag, self::$wp_sub_tags ) ) {
214	$this->in_sub_tag = substr( $tag, 3 );
215
216	return;
217	}
218
219	switch ( $tag ) {
220	case 'category':
221	if ( isset( $attr['domain'], $attr['nicename'] ) ) {
222	$this->sub_data['domain'] = $attr['domain'];
223	$this->sub_data['slug'] = $attr['nicename'];
224	}
225	break;
226	case 'item':
227	$this->in_post = true;
228	// No break !!!.
229	case 'title':
230	if ( $this->in_post ) {
231	$this->in_tag = 'post_title';
232	}
233	break;
234	case 'guid':
235	$this->in_tag = 'guid';
236	break;
237	case 'dc:creator':
238	$this->in_tag = 'post_author';
239	break;
240	case 'content:encoded':
241	$this->in_tag = 'post_content';
242	break;
243	case 'excerpt:encoded':
244	$this->in_tag = 'post_excerpt';
245	break;
246
247	case 'wp:term_slug':
248	$this->in_tag = 'slug';
249	break;
250	case 'wp:meta_key':
251	$this->in_sub_tag = 'key';
252	break;
253	case 'wp:meta_value':
254	$this->in_sub_tag = 'value';
255	break;
256	}
257	}
258
259	private function cdata( $cdata ) {
260	if ( ! trim( $cdata ) ) {
261	return;
262	}
263
264	if ( false !== $this->in_tag \|\| false !== $this->in_sub_tag ) {
265	$this->cdata .= $cdata;
266	} else {
267	$this->cdata .= trim( $cdata );
268	}
269	}
270
271	private function tag_close( $tag ) {
272	switch ( $tag ) {
273	case 'wp:comment':
274	unset( $this->sub_data['key'], $this->sub_data['value'] ); // Remove meta sub_data.
275	if ( ! empty( $this->sub_data ) ) {
276	$this->data['comments'][] = $this->sub_data;
277	}
278	$this->sub_data = [];
279	break;
280	case 'wp:commentmeta':
281	$this->sub_data['commentmeta'][] = [
282	'key' => $this->sub_data['key'],
283	'value' => $this->sub_data['value'],
284	];
285	break;
286	case 'category':
287	if ( ! empty( $this->sub_data ) ) {
288	$this->sub_data['name'] = $this->cdata;
289	$this->data['terms'][] = $this->sub_data;
290	}
291	$this->sub_data = [];
292	break;
293	case 'wp:postmeta':
294	if ( ! empty( $this->sub_data ) ) {
295	$this->data['postmeta'][] = $this->sub_data;
296	}
297	$this->sub_data = [];
298	break;
299	case 'item':
300	// Ensure attachment_type field is set, default to null if not present
301	if ( ! isset( $this->data['attachment_type'] ) ) {
302	$this->data['attachment_type'] = null;
303	}
304	$this->posts[] = $this->data;
305	$this->data = [];
306	break;
307	case 'wp:category':
308	case 'wp:tag':
309	case 'wp:term':
310	$n = substr( $tag, 3 );
311	array_push( $this->$n, $this->data );
312	$this->data = [];
313	break;
314	case 'wp:termmeta':
315	if ( ! empty( $this->sub_data ) ) {
316	$this->data['termmeta'][] = $this->sub_data;
317	}
318	$this->sub_data = [];
319	break;
320	case 'wp:author':
321	if ( ! empty( $this->data['author_login'] ) ) {
322	$this->authors[ $this->data['author_login'] ] = $this->data;
323	}
324	$this->data = [];
325	break;
326	case 'wp:base_site_url':
327	$this->base_url = $this->cdata;
328	if ( ! isset( $this->base_blog_url ) ) {
329	$this->base_blog_url = $this->cdata;
330	}
331	break;
332	case 'wp:base_blog_url':
333	$this->base_blog_url = $this->cdata;
334	break;
335	case 'wp:wxr_version':
336	$this->wxr_version = $this->cdata;
337	break;
338
339	default:
340	if ( $this->in_sub_tag ) {
341	$this->sub_data[ $this->in_sub_tag ] = $this->cdata;
342	$this->in_sub_tag = false;
343	} else if ( $this->in_tag ) {
344	$this->data[ $this->in_tag ] = $this->cdata;
345	$this->in_tag = false;
346	}
347	}
348
349	$this->cdata = '';
350	}
351
352	private function clear() {
353	$this->wxr_version = '';
354
355	$this->cdata = '';
356	$this->data = [];
357	$this->sub_data = [];
358
359	$this->in_post = false;
360	$this->in_tag = false;
361	$this->in_sub_tag = false;
362
363	$this->authors = [];
364	$this->posts = [];
365	$this->term = [];
366	$this->category = [];
367	$this->tag = [];
368	}
369	}
370