sitemap-buffer.php — Jetpack – WP Security, Backup, Speed, & Growth 14.8.1

jetpack / modules / sitemaps / sitemap-buffer.php

jetpack / modules / sitemaps Last commit date

sitemap-buffer.php

310 lines

1	<?php // phpcs:ignore WordPress.Files.FileName.InvalidClassFileName
2	/**
3	* Sitemaps (per the protocol) are essentially lists of XML fragments;
4	* lists which are subject to size constraints. The Jetpack_Sitemap_Buffer
5	* class abstracts the details of constructing these lists while
6	* maintaining the constraints.
7	*
8	* @since 4.8.0
9	* @package automattic/jetpack
10	*/
11
12	/**
13	* A buffer for constructing sitemap xml files.
14	*
15	* Models a list of strings such that
16	*
17	* 1. the list must have a bounded number of entries,
18	* 2. the concatenation of the strings must have bounded
19	* length (including some header and footer strings), and
20	* 3. each item has a timestamp, and we need to keep track
21	* of the most recent timestamp of the items in the list.
22	*
23	* @since 4.8.0
24	*/
25	abstract class Jetpack_Sitemap_Buffer {
26
27	/**
28	* Largest number of items the buffer can hold.
29	*
30	* @access protected
31	* @since 4.8.0
32	* @var int $item_capacity The item capacity.
33	*/
34	protected $item_capacity;
35
36	/**
37	* Largest number of bytes the buffer can hold.
38	*
39	* @access protected
40	* @since 4.8.0
41	* @var int $byte_capacity The byte capacity.
42	*/
43	protected $byte_capacity;
44
45	/**
46	* Flag which detects when the buffer is full.
47	*
48	* @access protected
49	* @since 4.8.0
50	* @var bool $is_full_flag The flag value. This flag is set to false on construction and only flipped to true if we've tried to add something and failed.
51	*/
52	protected $is_full_flag;
53
54	/**
55	* Flag which detects when the buffer is empty.
56	*
57	* @access protected
58	* @since 4.8.0
59	* @var bool $is_empty_flag The flag value. This flag is set to true on construction and only flipped to false if we've tried to add something and succeeded.
60	*/
61	protected $is_empty_flag;
62
63	/**
64	* The most recent timestamp seen by the buffer.
65	*
66	* @access protected
67	* @since 4.8.0
68	* @var string $timestamp Must be in 'YYYY-MM-DD hh:mm:ss' format.
69	*/
70	protected $timestamp;
71
72	/**
73	* The DOM document object that is currently being used to construct the XML doc.
74	*
75	* @access protected
76	* @since 5.3.0
77	* @var DOMDocument $doc
78	*/
79	protected $doc = null;
80
81	/**
82	* The root DOM element object that holds everything inside. Do not use directly, call
83	* the get_root_element getter method instead.
84	*
85	* @access protected
86	* @since 5.3.0
87	* @var DOMElement $doc
88	*/
89	protected $root = null;
90
91	/**
92	* Helper class to construct sitemap paths.
93	*
94	* @since 5.3.0
95	* @protected
96	* @var Jetpack_Sitemap_Finder
97	*/
98	protected $finder;
99
100	/**
101	* Construct a new Jetpack_Sitemap_Buffer.
102	*
103	* @since 4.8.0
104	*
105	* @param int $item_limit The maximum size of the buffer in items.
106	* @param int $byte_limit The maximum size of the buffer in bytes.
107	* @param string $time The initial datetime of the buffer. Must be in 'YYYY-MM-DD hh:mm:ss' format.
108	*/
109	public function __construct( $item_limit, $byte_limit, $time ) {
110	$this->is_full_flag = false;
111	$this->timestamp = $time;
112
113	$this->finder = new Jetpack_Sitemap_Finder();
114	$this->doc = new DOMDocument( '1.0', 'UTF-8' );
115	$this->doc->formatOutput = true;
116	$this->doc->preserveWhiteSpace = false;
117
118	$this->item_capacity = max( 1, (int) $item_limit );
119	$this->byte_capacity = max( 1, (int) $byte_limit ) - strlen( $this->doc->saveXML() );
120	}
121
122	/**
123	* Returns a DOM element that contains all sitemap elements.
124	*
125	* @access protected
126	* @since 5.3.0
127	* @return DOMElement $root
128	*/
129	abstract protected function get_root_element();
130
131	/**
132	* Append an item to the buffer, if there is room for it,
133	* and set is_empty_flag to false. If there is no room,
134	* we set is_full_flag to true. If $item is null,
135	* don't do anything and report success.
136	*
137	* @since 5.3.0
138	*
139	* @param array $array The item to be added.
140	*
141	* @return bool True if the append succeeded, False if not.
142	*/
143	public function append( $array ) {
144	if ( $array === null ) {
145	return true;
146	}
147
148	if ( $this->is_full_flag ) {
149	return false;
150	}
151
152	if ( 0 >= $this->item_capacity \|\| 0 >= $this->byte_capacity ) {
153	$this->is_full_flag = true;
154	return false;
155	} else {
156	$this->item_capacity -= 1;
157	$added_element = $this->array_to_xml_string( $array, $this->get_root_element(), $this->doc );
158
159	$this->byte_capacity -= strlen( $this->doc->saveXML( $added_element ) );
160
161	return true;
162	}
163	}
164
165	/**
166	* Retrieve the contents of the buffer.
167	*
168	* @since 4.8.0
169	*
170	* @return string The contents of the buffer (with the footer included).
171	*/
172	public function contents() {
173	if ( $this->is_empty() ) {
174	// The sitemap should have at least the root element added to the DOM.
175	$this->get_root_element();
176	}
177	return $this->doc->saveXML();
178	}
179
180	/**
181	* Retrieve the document object.
182	*
183	* @since 5.3.0
184	* @return DOMDocument $doc
185	*/
186	public function get_document() {
187	return $this->doc;
188	}
189
190	/**
191	* Detect whether the buffer is full.
192	*
193	* @since 4.8.0
194	*
195	* @return bool True if the buffer is full, false otherwise.
196	*/
197	public function is_full() {
198	return $this->is_full_flag;
199	}
200
201	/**
202	* Detect whether the buffer is empty.
203	*
204	* @since 4.8.0
205	*
206	* @return bool True if the buffer is empty, false otherwise.
207	*/
208	public function is_empty() {
209	return (
210	! isset( $this->root )
211	\|\| ! $this->root->hasChildNodes()
212	);
213	}
214
215	/**
216	* Update the timestamp of the buffer.
217	*
218	* @since 4.8.0
219	*
220	* @param string $new_time A datetime string in 'YYYY-MM-DD hh:mm:ss' format.
221	*/
222	public function view_time( $new_time ) {
223	$this->timestamp = max( $this->timestamp, $new_time );
224	}
225
226	/**
227	* Retrieve the timestamp of the buffer.
228	*
229	* @since 4.8.0
230	*
231	* @return string A datetime string in 'YYYY-MM-DD hh:mm:ss' format.
232	*/
233	public function last_modified() {
234	return $this->timestamp;
235	}
236
237	/**
238	* Render an associative array as an XML string. This is needed because
239	* SimpleXMLElement only handles valid XML, but we sometimes want to
240	* pass around (possibly invalid) fragments. Note that 'null' values make
241	* a tag self-closing; this is only sometimes correct (depending on the
242	* version of HTML/XML); see the list of 'void tags'.
243	*
244	* Example:
245	*
246	* array(
247	* 'html' => array( \|<html xmlns="foo">
248	* 'head' => array( \| <head>
249	* 'title' => 'Woo!', \| <title>Woo!</title>
250	* ), \| </head>
251	* 'body' => array( ==> \| <body>
252	* 'h2' => 'Some thing', \| <h2>Some thing</h2>
253	* 'p' => 'it's all up ons', \| <p>it's all up ons</p>
254	* 'br' => null, \| <br />
255	* ), \| </body>
256	* ), \|</html>
257	* )
258	*
259	* @access protected
260	* @since 3.9.0
261	* @since 4.8.0 Rename, add $depth parameter, and change return type.
262	* @since 5.3.0 Refactor, remove $depth parameter, add $parent and $root, make access protected.
263	*
264	* @param array $array A recursive associative array of tag/child relationships.
265	* @param DOMElement $parent (optional) an element to which new children should be added.
266	* @param DOMDocument $root (optional) the parent document.
267	*
268	* @return string\|DOMDocument The rendered XML string or an object if root element is specified.
269	*/
270	protected function array_to_xml_string( $array, $parent = null, $root = null ) {
271	$element = null;
272	$return_string = false;
273
274	if ( null === $parent ) {
275	$return_string = true;
276	$root = new DOMDocument();
277	$parent = $root;
278	}
279
280	if ( is_array( $array ) ) {
281
282	foreach ( $array as $key => $value ) {
283	$element = $root->createElement( $key );
284	$parent->appendChild( $element );
285
286	if ( is_array( $value ) ) {
287	foreach ( $value as $child_key => $child_value ) {
288	$child = $root->createElement( $child_key );
289	$element->appendChild( $child );
290	$child->appendChild( self::array_to_xml_string( $child_value, $child, $root ) );
291	}
292	} else {
293	$element->appendChild(
294	$root->createTextNode( $value )
295	);
296	}
297	}
298	} else {
299	$element = $root->createTextNode( $array );
300	$parent->appendChild( $element );
301	}
302
303	if ( $return_string ) {
304	return $root->saveHTML();
305	} else {
306	return $element;
307	}
308	}
309	}
310