betterdocs / includes / Admin / Importer / Parsers / WXR_Parser_XML.php

WXR_Parser_XML.php in BetterDocs – AI Documentation, Knowledge Base, MCP Server, Docs, Wikis, FAQ & Chatbot 4.9.3, at includes/Admin/Importer/Parsers/WXR_Parser_XML.php

372 lines 7.6 KB

Raw Download Zip

1	<?php
2	namespace WPDeveloper\BetterDocs\Admin\Importer\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13	* Originally made by WordPress part of WordPress/Importer.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:status',
37	'wp:post_name',
38	'wp:post_parent',
39	'wp:menu_order',
40	'wp:post_type',
41	'wp:post_password',
42	'wp:is_sticky',
43	'wp:term_id',
44	'wp:category_nicename',
45	'wp:category_parent',
46	'wp:cat_name',
47	'wp:category_description',
48	'wp:tag_slug',
49	'wp:tag_name',
50	'wp:tag_description',
51	'wp:term_taxonomy',
52	'wp:term_parent',
53	'wp:term_name',
54	'wp:term_description',
55	'wp:author_id',
56	'wp:author_login',
57	'wp:author_email',
58	'wp:author_display_name',
59	'wp:author_first_name',
60	'wp:author_last_name',
61	];
62
63	private static $wp_sub_tags = [
64	'wp:comment_id',
65	'wp:comment_author',
66	'wp:comment_author_email',
67	'wp:comment_author_url',
68	'wp:comment_author_IP',
69	'wp:comment_date',
70	'wp:comment_date_gmt',
71	'wp:comment_content',
72	'wp:comment_approved',
73	'wp:comment_type',
74	'wp:comment_parent',
75	'wp:comment_user_id',
76	];
77
78	/**
79	* @var string
80	*/
81	private $wxr_version;
82
83	/**
84	* @var string
85	*/
86	private $cdata;
87
88	/**
89	* @var array
90	*/
91	private $data;
92
93	/**
94	* @var array
95	*/
96	private $sub_data;
97
98	/**
99	* @var boolean
100	*/
101	private $in_post;
102
103	/**
104	* @var boolean
105	*/
106	private $in_tag;
107
108	/**
109	* @var boolean
110	*/
111	private $in_sub_tag;
112
113	/**
114	* @var array
115	*/
116	private $authors;
117
118	/**
119	* @var array
120	*/
121	private $posts;
122
123	/**
124	* @var array
125	*/
126	private $term;
127
128	/**
129	* @var array
130	*/
131	private $category;
132
133	/**
134	* @var array
135	*/
136	private $tag;
137
138	/**
139	* @var string
140	*/
141	private $base_url;
142
143	/**
144	* @var string
145	*/
146	private $base_blog_url;
147
148	/**
149	* @param string $file
150	*
151	* @return array\|WP_Error
152	*/
153	public function parse( $file ) {
154	$this->clear();
155
156	$xml = xml_parser_create( 'UTF-8' );
157	xml_parser_set_option( $xml, XML_OPTION_SKIP_WHITE, 1 );
158	xml_parser_set_option( $xml, XML_OPTION_CASE_FOLDING, 0 );
159	xml_set_object( $xml, $this );
160
161	xml_set_character_data_handler(
162	$xml,
163	function ( $parser, $cdata ) {
164	$this->cdata( $cdata );
165	}
166	);
167
168	$tag_open_callback = function ( $parse, $tag, $attr ) {
169	$this->tag_open( $tag, $attr );
170	};
171
172	$tag_close_callback = function ( $parser, $tag ) {
173	$this->tag_close( $tag );
174	};
175
176	xml_set_element_handler( $xml, $tag_open_callback, $tag_close_callback );
177
178	if ( ! xml_parse( $xml, Utils::file_get_contents( $file ), true ) ) {
179	$current_line = xml_get_current_line_number( $xml );
180	$current_column = xml_get_current_column_number( $xml );
181	$error_code = xml_get_error_code( $xml );
182	$error_string = xml_error_string( $error_code );
183
184	return new WP_Error(
185	'XML_parse_error',
186	'There was an error when reading this WXR file',
187	[
188	$current_line,
189	$current_column,
190	$error_string,
191	]
192	);
193	}
194	xml_parser_free( $xml );
195
196	if ( ! preg_match( '/^\d+\.\d+$/', $this->wxr_version ) ) {
197	return new WP_Error( 'WXR_parse_error', esc_html__( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'betterdocs' ) );
198	}
199
200	return array(
201	'authors' => $this->authors,
202	'posts' => $this->posts,
203	'categories' => $this->category,
204	'tags' => $this->tag,
205	'terms' => $this->term,
206	'base_url' => $this->base_url,
207	'base_blog_url' => $this->base_blog_url,
208	'version' => $this->wxr_version,
209	);
210	}
211
212	private function tag_open( $tag, $attr ) {
213	if ( in_array( $tag, self::$wp_tags ) ) {
214	$this->in_tag = substr( $tag, 3 );
215
216	return;
217	}
218
219	if ( in_array( $tag, self::$wp_sub_tags ) ) {
220	$this->in_sub_tag = substr( $tag, 3 );
221
222	return;
223	}
224
225	switch ( $tag ) {
226	case 'category':
227	if ( isset( $attr['domain'], $attr['nicename'] ) ) {
228	$this->sub_data['domain'] = $attr['domain'];
229	$this->sub_data['slug'] = $attr['nicename'];
230	}
231	break;
232	case 'item':
233	$this->in_post = true;
234	// No break !!!.
235	case 'title':
236	if ( $this->in_post ) {
237	$this->in_tag = 'post_title';
238	}
239	break;
240	case 'guid':
241	$this->in_tag = 'guid';
242	break;
243	case 'dc:creator':
244	$this->in_tag = 'post_author';
245	break;
246	case 'content:encoded':
247	$this->in_tag = 'post_content';
248	break;
249	case 'excerpt:encoded':
250	$this->in_tag = 'post_excerpt';
251	break;
252
253	case 'wp:term_slug':
254	$this->in_tag = 'slug';
255	break;
256	case 'wp:meta_key':
257	$this->in_sub_tag = 'key';
258	break;
259	case 'wp:meta_value':
260	$this->in_sub_tag = 'value';
261	break;
262	}
263	}
264
265	private function cdata( $cdata ) {
266	if ( ! trim( $cdata ) ) {
267	return;
268	}
269
270	if ( false !== $this->in_tag \|\| false !== $this->in_sub_tag ) {
271	$this->cdata .= $cdata;
272	} else {
273	$this->cdata .= trim( $cdata );
274	}
275	}
276
277	private function tag_close( $tag ) {
278	switch ( $tag ) {
279	case 'wp:comment':
280	unset( $this->sub_data['key'], $this->sub_data['value'] ); // Remove meta sub_data.
281	if ( ! empty( $this->sub_data ) ) {
282	$this->data['comments'][] = $this->sub_data;
283	}
284	$this->sub_data = [];
285	break;
286	case 'wp:commentmeta':
287	$this->sub_data['commentmeta'][] = [
288	'key' => $this->sub_data['key'],
289	'value' => $this->sub_data['value'],
290	];
291	break;
292	case 'category':
293	if ( ! empty( $this->sub_data ) ) {
294	$this->sub_data['name'] = $this->cdata;
295	$this->data['terms'][] = $this->sub_data;
296	}
297	$this->sub_data = [];
298	break;
299	case 'wp:postmeta':
300	if ( ! empty( $this->sub_data ) ) {
301	$this->data['postmeta'][] = $this->sub_data;
302	}
303	$this->sub_data = [];
304	break;
305	case 'item':
306	$this->posts[] = $this->data;
307	$this->data = [];
308	break;
309	case 'wp:category':
310	case 'wp:tag':
311	case 'wp:term':
312	$n = substr( $tag, 3 );
313	array_push( $this->$n, $this->data );
314	$this->data = [];
315	break;
316	case 'wp:termmeta':
317	if ( ! empty( $this->sub_data ) ) {
318	$this->data['termmeta'][] = $this->sub_data;
319	}
320	$this->sub_data = [];
321	break;
322	case 'wp:author':
323	if ( ! empty( $this->data['author_login'] ) ) {
324	$this->authors[ $this->data['author_login'] ] = $this->data;
325	}
326	$this->data = [];
327	break;
328	case 'wp:base_site_url':
329	$this->base_url = $this->cdata;
330	if ( ! isset( $this->base_blog_url ) ) {
331	$this->base_blog_url = $this->cdata;
332	}
333	break;
334	case 'wp:base_blog_url':
335	$this->base_blog_url = $this->cdata;
336	break;
337	case 'wp:wxr_version':
338	$this->wxr_version = $this->cdata;
339	break;
340
341	default:
342	if ( $this->in_sub_tag ) {
343	$this->sub_data[ $this->in_sub_tag ] = $this->cdata;
344	$this->in_sub_tag = false;
345	} elseif ( $this->in_tag ) {
346	$this->data[ $this->in_tag ] = $this->cdata;
347	$this->in_tag = false;
348	}
349	}
350
351	$this->cdata = '';
352	}
353
354	private function clear() {
355	$this->wxr_version = '';
356
357	$this->cdata = '';
358	$this->data = [];
359	$this->sub_data = [];
360
361	$this->in_post = false;
362	$this->in_tag = false;
363	$this->in_sub_tag = false;
364
365	$this->authors = [];
366	$this->posts = [];
367	$this->term = [];
368	$this->category = [];
369	$this->tag = [];
370	}
371	}
372