windpress / vendor / masterminds / html5 / src / HTML5.php

HTML5.php in WindPress – Tailwind CSS integration for WordPress 3.2.88, at vendor/masterminds/html5/src/HTML5.php

222 lines 8.4 KB

Raw Download Zip

1	<?php
2
3	namespace WindPressDeps\Masterminds;
4
5	use WindPressDeps\Masterminds\HTML5\Parser\DOMTreeBuilder;
6	use WindPressDeps\Masterminds\HTML5\Parser\Scanner;
7	use WindPressDeps\Masterminds\HTML5\Parser\Tokenizer;
8	use WindPressDeps\Masterminds\HTML5\Serializer\OutputRules;
9	use WindPressDeps\Masterminds\HTML5\Serializer\Traverser;
10	/**
11	* This class offers convenience methods for parsing and serializing HTML5.
12	* It is roughly designed to mirror the \DOMDocument native class.
13	*/
14	class HTML5
15	{
16	/**
17	* Global options for the parser and serializer.
18	*
19	* @var array
20	*/
21	private $defaultOptions = array(
22	// Whether the serializer should aggressively encode all characters as entities.
23	'encode_entities' => \false,
24	// Prevents the parser from automatically assigning the HTML5 namespace to the DOM document.
25	'disable_html_ns' => \false,
26	);
27	protected $errors = array();
28	public function __construct(array $defaultOptions = array())
29	{
30	$this->defaultOptions = array_merge($this->defaultOptions, $defaultOptions);
31	}
32	/**
33	* Get the current default options.
34	*
35	* @return array
36	*/
37	public function getOptions()
38	{
39	return $this->defaultOptions;
40	}
41	/**
42	* Load and parse an HTML file.
43	*
44	* This will apply the HTML5 parser, which is tolerant of many
45	* varieties of HTML, including XHTML 1, HTML 4, and well-formed HTML
46	* 3. Note that in these cases, not all of the old data will be
47	* preserved. For example, XHTML's XML declaration will be removed.
48	*
49	* The rules governing parsing are set out in the HTML 5 spec.
50	*
51	* @param string\|resource $file The path to the file to parse. If this is a resource, it is
52	* assumed to be an open stream whose pointer is set to the first
53	* byte of input.
54	* @param array $options Configuration options when parsing the HTML.
55	*
56	* @return \DOMDocument A DOM document. These object type is defined by the libxml
57	* library, and should have been included with your version of PHP.
58	*/
59	public function load($file, array $options = array())
60	{
61	// Handle the case where file is a resource.
62	if (is_resource($file)) {
63	return $this->parse(stream_get_contents($file), $options);
64	}
65	return $this->parse(file_get_contents($file), $options);
66	}
67	/**
68	* Parse a HTML Document from a string.
69	*
70	* Take a string of HTML 5 (or earlier) and parse it into a
71	* DOMDocument.
72	*
73	* @param string $string A html5 document as a string.
74	* @param array $options Configuration options when parsing the HTML.
75	*
76	* @return \DOMDocument A DOM document. DOM is part of libxml, which is included with
77	* almost all distribtions of PHP.
78	*/
79	public function loadHTML($string, array $options = array())
80	{
81	return $this->parse($string, $options);
82	}
83	/**
84	* Convenience function to load an HTML file.
85	*
86	* This is here to provide backwards compatibility with the
87	* PHP DOM implementation. It simply calls load().
88	*
89	* @param string $file The path to the file to parse. If this is a resource, it is
90	* assumed to be an open stream whose pointer is set to the first
91	* byte of input.
92	* @param array $options Configuration options when parsing the HTML.
93	*
94	* @return \DOMDocument A DOM document. These object type is defined by the libxml
95	* library, and should have been included with your version of PHP.
96	*/
97	public function loadHTMLFile($file, array $options = array())
98	{
99	return $this->load($file, $options);
100	}
101	/**
102	* Parse a HTML fragment from a string.
103	*
104	* @param string $string the HTML5 fragment as a string
105	* @param array $options Configuration options when parsing the HTML
106	*
107	* @return \DOMDocumentFragment A DOM fragment. The DOM is part of libxml, which is included with
108	* almost all distributions of PHP.
109	*/
110	public function loadHTMLFragment($string, array $options = array())
111	{
112	return $this->parseFragment($string, $options);
113	}
114	/**
115	* Return all errors encountered into parsing phase.
116	*
117	* @return array
118	*/
119	public function getErrors()
120	{
121	return $this->errors;
122	}
123	/**
124	* Return true it some errors were encountered into parsing phase.
125	*
126	* @return bool
127	*/
128	public function hasErrors()
129	{
130	return count($this->errors) > 0;
131	}
132	/**
133	* Parse an input string.
134	*
135	* @param string $input
136	*
137	* @return \DOMDocument
138	*/
139	public function parse($input, array $options = array())
140	{
141	$this->errors = array();
142	$options = array_merge($this->defaultOptions, $options);
143	$events = new DOMTreeBuilder(\false, $options);
144	$scanner = new Scanner($input, !empty($options['encoding']) ? $options['encoding'] : 'UTF-8');
145	$parser = new Tokenizer($scanner, $events, !empty($options['xmlNamespaces']) ? Tokenizer::CONFORMANT_XML : Tokenizer::CONFORMANT_HTML);
146	$parser->parse();
147	$this->errors = $events->getErrors();
148	return $events->document();
149	}
150	/**
151	* Parse an input stream where the stream is a fragment.
152	*
153	* Lower-level loading function. This requires an input stream instead
154	* of a string, file, or resource.
155	*
156	* @param string $input The input data to parse in the form of a string.
157	* @param array $options An array of options.
158	*
159	* @return \DOMDocumentFragment
160	*/
161	public function parseFragment($input, array $options = array())
162	{
163	$options = array_merge($this->defaultOptions, $options);
164	$events = new DOMTreeBuilder(\true, $options);
165	$scanner = new Scanner($input, !empty($options['encoding']) ? $options['encoding'] : 'UTF-8');
166	$parser = new Tokenizer($scanner, $events, !empty($options['xmlNamespaces']) ? Tokenizer::CONFORMANT_XML : Tokenizer::CONFORMANT_HTML);
167	$parser->parse();
168	$this->errors = $events->getErrors();
169	return $events->fragment();
170	}
171	/**
172	* Save a DOM into a given file as HTML5.
173	*
174	* @param mixed $dom The DOM to be serialized.
175	* @param string\|resource $file The filename to be written or resource to write to.
176	* @param array $options Configuration options when serializing the DOM. These include:
177	* - encode_entities: Text written to the output is escaped by default and not all
178	* entities are encoded. If this is set to true all entities will be encoded.
179	* Defaults to false.
180	*/
181	public function save($dom, $file, $options = array())
182	{
183	$close = \true;
184	if (is_resource($file)) {
185	$stream = $file;
186	$close = \false;
187	} else {
188	$stream = fopen($file, 'wb');
189	}
190	$options = array_merge($this->defaultOptions, $options);
191	$rules = new OutputRules($stream, $options);
192	$trav = new Traverser($dom, $stream, $rules, $options);
193	$trav->walk();
194	/*
195	* release the traverser to avoid cyclic references and allow PHP to free memory without waiting for gc_collect_cycles
196	*/
197	$rules->unsetTraverser();
198	if ($close) {
199	fclose($stream);
200	}
201	}
202	/**
203	* Convert a DOM into an HTML5 string.
204	*
205	* @param mixed $dom The DOM to be serialized.
206	* @param array $options Configuration options when serializing the DOM. These include:
207	* - encode_entities: Text written to the output is escaped by default and not all
208	* entities are encoded. If this is set to true all entities will be encoded.
209	* Defaults to false.
210	*
211	* @return string A HTML5 documented generated from the DOM.
212	*/
213	public function saveHTML($dom, $options = array())
214	{
215	$stream = fopen('php://temp', 'wb');
216	$this->save($dom, $stream, array_merge($this->defaultOptions, $options));
217	$html = stream_get_contents($stream, -1, 0);
218	fclose($stream);
219	return $html;
220	}
221	}
222